Assessment of Age-Related Differences in Lower Leg Muscles Quality Using Radiomic Features of Magnetic Resonance Images

Takuro Shiiba,Suzumi Mori,Takuya Shimozono,Shuji Ito,Kazuki Takano

DOI: https://doi.org/10.1007/s10278-024-01268-7

2024-09-16

Abstract:Sarcopenia, characterised by a decline in muscle mass and strength, affects the health of the elderly, leading to increased falls, hospitalisation, and mortality rates. Muscle quality, reflecting microscopic and macroscopic muscle changes, is a critical determinant of physical function. To utilise radiomic features extracted from magnetic resonance (MR) images to assess age-related changes in muscle quality, a dataset of 24 adults, divided into older (male/female: 6/6, 66-79 years) and younger (male/female: 6/6, 21-31 years) groups, was used to investigate the radiomics features of the dorsiflexor and plantar flexor muscles of the lower leg that are critical for mobility. MR images were processed using MaZda software for feature extraction. Dimensionality reduction was performed using principal component analysis and recursive feature elimination, followed by classification using machine learning models, such as support vector machine, extreme gradient boosting, and naïve Bayes. A leave-one-out validation test was used to train and test the classifiers, and the area under the receiver operating characteristic curve (AUC) was used to evaluate the classification performance. The analysis revealed that significant differences in radiomic feature distributions were found between age groups, with older adults showing higher complexity and variability in muscle texture. The plantar flexors showed similar or higher AUC than the dorsiflexors in all models. When the dorsiflexor muscles were combined with the plantar flexor muscles, they tended to have a higher AUC than when they were used alone. Radiomic features in lower-leg MR images reflect ageing, especially in the plantar flexor muscles. Radiomic analysis can offer a deeper understanding of age-related muscle quality than traditional muscle mass assessments.